Partnering opportunity triage, explained.

Partnering opportunity triage explained: reading every pitch once, and answering every sender

One pitch, from the sender's outbox to a signed licensing deal

1. The pitch arrivesThrough a portal, an inbox, a banker or a meeting. Each channel carries a different document and a different level of detail.
2. Triage reads, scores and answers itHours of expert time at most, in the first days. Keep it closed if marked confidential, find any earlier pitch, score fit, route or decline, and reply.
3. A reviewer forms a viewA therapeutic-area scientist reads the cited summary and asks for a call, more data, or nothing further.
4. Both sides sign a confidentiality agreementOnly now does confidential data move, usually into a data room (a secure online folder), for weeks or months of detailed checking, called diligence.
5. Term sheet, then contractThe deal itself: a short paper of main terms, then the full contract. One biotech licensing lead says a license from a large company took "the better part of a year" to complete.3

Step 2 is the subject of this guide. Steps 4 and 5 are covered only where triage touches them.

Most pitches end at the first reader, and three more facts about the front door

Search and evaluation, non-confidential deck, CDA, search criteria: the terms you meet first

Fourteen terms are defined in the one-page summary.

What search and evaluation decides first: five answers a pitch can get

In shortTriage gives one of five answers: route to a reviewer, take a second look, ask for more, decline, or hold for legal. It decides who spends time on what; the deal decision comes later, from others. The pool it draws from is enormous, and the share that survives is small.

The pipeline of medicines in development worldwide numbered about 22,940 in Citeline's 2026 count, the first fall since the mid-1990s.6 Many of those assets have owners who would listen to an offer. Search teams go looking for some of them. Others arrive on their own, and those are the inbound flow. A GSK business development leader told the same 2024 panel, "It's pretty rare we don't know about something".1 So a pitch often lands on someone who has heard of the asset already and cannot find what the company concluded last time.

Read that funnel carefully. Most of the drop, 280 of 400, happened before anyone signed anything, on the strength of a story told in an hour-long call. That is the triage stage. A company that treats it as clerical has handed its largest decision to whoever had time that week.

Route, second look, ask for more, decline, hold: what each answer asks of the reader

How many conference meetings feed the partnering inbox

Partnering conferences show how the flow has grown. BIO-Europe Spring recorded 4,001 one-to-one meetings in 2007 and 17,338 in 2026, and its peak was 19,340 in 2024.7 The BIO International Convention reports 66,308 partnering meetings for 2025.8 Each meeting can leave a page of notes on someone's laptop, and those notes are pitches too.

The law around an inbound pitch, and what changed for China-origin assets in 2025 and 2026

In shortNo regulator writes rules for triage. Trade secret law, securities law, contract law, merger filings and national security rules all reach it anyway. The news is Treasury's October 2026 proposal to bring biotechnology into its outbound investment program, which restricts American investment in sensitive Chinese technology. Law firms read it as leaving pure licensing outside, for now. It is a proposal, comments close December 7, 2026, and a bill would go further.

What Treasury's October 2026 proposal does and does not say about licensing

Two quick readings of this proposal are both wrong: that biotechnology sits outside the outbound rules, and that licensing from China is now restricted. Treasury's proposed rule, published October 26, 2026, does add biotechnology to the outbound program. It covers using a biological system to produce or modify, among other things, "a pharmaceutical compound", including small molecules, vaccines, cell and gene therapies and monoclonal antibodies.15

We read the proposal's definitions and found no category for licensing as a transaction. Law firms agree, in careful words, and their reading turns on the shape of the deal.

Three things could change this. Treasury is taking comments to December 7, 2026, and asks whether and how "pharmaceutical compounds" should be "more precisely scoped to focus on advanced or frontier technologies".15 BINSA would reach licensing directly.13,14 And a committee chair said on October 23, 2026 that he would work on how the final rule treats dependence on Chinese biotech assets.30 A separate draft executive order on Chinese licensing deals was reported in September 2025, and a White House spokesperson denied "actively considering" it; we found no signed order.31

What this means for a triage reader on a Tuesday

Counsel decides these questions. Triage records the facts that let counsel decide them quickly: where the company is domiciled, who owns it, who manufactures the drug, where the human data were generated, and whether the ask includes equity or a joint venture. A field left blank here becomes a week lost later.

For a China-origin asset, the same record feeds the reviewer's scientific question. FDA's advisory committee voted 14 to 1 in February 2022 against a drug supported by a trial run only in China, and said the sponsors should run trials reflecting US patients.9 A pitch whose data come from one country is a pitch with a question attached.

Portal form, BD inbox, banker teaser, meeting notes: what each channel gives the first reader

In shortA pitch comes in four shapes, and they differ in completeness and in urgency. A portal form is tidy and thin. A deck is full and unstructured. A banker's teaser is short and comes with a bid date. Meeting notes are half a page written from memory. A reader's first task is to turn all four into the same set of fields, and to leave a field empty when the pitch did not say.

Partner pages tell senders what to send, in few words. One large device and diagnostics company asks for "nonconfidential information" and points to a partnering portal.34 A drug company's page says it assesses every project against one published framework so it can give "consistent, objective feedback".35 The response time, on every page we could read, goes unstated.

Eleven fields that turn any pitch into one line of a register, the team's running list of pitches

Teams differ on the details, and most registers converge on the same short list. The discipline that matters is the citation. Each value carries the slide or page it came from, because the reviewer will not trust a summary she cannot check, and the company may be misquoted otherwise.

Why a file marked confidential must stay closed until a CDA is signed

In shortA large company's scientists may already be working on something close to what a stranger sends. If they read the stranger's secret, a later patent or program can be challenged as copied. So the check comes before reading, it looks only at the cover and the file labels, and it logs a receipt that proves the file stayed shut.

Put yourself in the recipient's position. Your oncology group has an unpublished idea for a new cancer drug. An email arrives with an attachment called "confidential data package", describing something similar. If anyone opens it, you now have two problems. The sender may later say you took its idea, and you may have to prove you did not.

The Defend Trade Secrets Act offers a defense, since independent derivation is not an improper means of acquiring a secret.16 That defense is much easier to run if you can show nobody read the file.

One consumer company's portal states the logic plainly: taking only non-confidential material "eliminates the risk of jeopardizing" its own research, and a confidentiality agreement comes only if the overview is of interest.5 Drug companies follow the same order, and a reply to a confidential sender says so, offers a CDA, and asks for a non-confidential summary in the meantime.

Four ways the confidential boundary fails at a real front door

What a CDA should say before a pitch can move on to data

Once both sides want to go further, counsel's questions are about scope and memory. How long does the obligation last? Does it cover the fact of the talks? And is there a residuals clause, which lets the recipient keep using what its people remember? Farrer & Co, a law firm, notes that such clauses are common in agreements written by large technology and pharma companies, and warns that they can let a recipient use remembered information at will.36

A small biotech should read that clause first, and a large company should know its own form contains it. Term lengths vary by deal; we found no public benchmark for how long these agreements take to negotiate, so measure yours from request to signature.

How a fit score works on an inbound pitch: six points, worked through with numbers

In shortA fit score asks one question: does this pitch match what this area said it wants, today? It is a weighted sum, and the arithmetic is simple. The score sorts a queue, and valuing the asset stays with the reviewers. It hides its own assumptions unless each point carries a reason and the criteria carry a version.

Here is a common way to build one. Choose a handful of points and give each a weight that sums to 100. For each pitch, a reader judges how well it meets each point, from 0 to 1, and multiplies. Take the example from the top of this guide, scored against an immunology area's criteria:

Next, cut-offs. A team might route at 65 or above, take a second look from 50 to 64, and decline below 50. Those lines are choices, and they should be tested against history. Score last year's pitches on the current rule and see which of them the reviewers kept. If half of what they kept scored under 65, the weights are wrong, whatever they look like on paper.

Six points that separate a fit score from a gut feeling

These six and their weights are one reasonable set, used in this guide's example. Yours should reflect your own strategy.

Reading a small study without being fooled by it

A deck's lead number deserves one quick test. QH-1187's slide reports clinical remission at week 12 in 33% of treated patients (6 of 18) against 7% on placebo (1 of 14). Those are encouraging numbers from a small study. A 95% confidence interval gives the range of true rates the data are consistent with, and here the two ranges overlap. Fisher's exact test asks how likely a gap at least this big would be if the drug did nothing.

The result earns the deck a reviewer's attention, and it is too thin to support a price.

The same care applies to the rest of the clinical data on a slide: how many patients, compared with what, measured when, and chosen how. A first reader records those four things and leaves the judgment to the reviewer.

What a phase is worth: published success rates, by therapeutic area

Drugs pass through three trial phases in people. Phase 1 looks at safety and dose in a small group, Phase 2 starts to test whether the drug works in people with the disease, and Phase 3 tests that in a much larger group.37

A fit score says nothing about the odds of getting through them, so readers should know the base rates. BIO, Informa and QLS Advisors studied 9,704 development programs from 2011 to 2020. Across all diseases, 7.9% of drugs entering Phase 1 reached approval, and Phase 2 was the hardest step, at 28.9%.38 Multiply the published transition rates and you get the chance from the start of Phase 2 to approval.

What a median upfront asks you to believe about an early-stage drug

Now put odds and price together. J.P. Morgan's report on DealForma data gives median upfront cash and equity for big pharma in-licensing in 2025: $125 million at Phase 1 on 14 deals, $180 million at Phase 2 on 8 deals and $240 million at Phase 3 on 6 deals.39 Divide each by the published chance of approval from that phase, and you get the value the drug must be worth on approval, in present value after all remaining costs, just to repay the upfront.

This model belongs to the reviewer and the valuation team, who use risk-adjusted net present value: a drug's future cash, discounted and weighted by its odds. A triage reader needs the sense of scale, so that an unrealistic ask is recorded as such. The upfront is usually a small part of an announced deal: 6% of announced licensing value in 2026 to date, on J.P. Morgan's figures.39 The example's $75 million on $1.3 billion is 5.8%.

The same molecule under a new name: how to recognize a re-pitched asset

In shortAn asset can come back with a new code, a new company name, a banker in front of it or a new disease. Each change breaks a search on the name. Asset identity lives in things that are harder to change: the targets and format, the manufacturer, the contact and the company's own history. A match should state its reasons, bring back the earlier decision and what has changed, and wait for a person.

Companies rename themselves after a funding round or a merger. Assets get a new code when a program moves to a new owner. A banker's teaser may hide the company's name until a CDA is signed. A molecule declined for one disease returns for another. These are ordinary business changes, and the sender may not know you saw it. What suffers is your own memory: a reviewer reads it again from the beginning, or a different reviewer says something different, and the company notices.

The earlier answer matters more than most teams assume. The royalty investor whose funnel opened this guide describes deliberately telling companies "it's too early. Why don't we meet in a year or two", then checking whether management did what it said.4 A watch list is a promise to remember, and it pays to keep it. When the condition you named has been met, the reply can say so in a line.

Three kinds of match that deserve three different answers

No public figure exists for how often assets are re-pitched. Take your last 100 pitches, search each by target and company history, and count.

Keeping a reply promise: how, in an illustrative queue, 90 minutes of work takes three weeks

In shortReplies are late because pitches wait. Queue arithmetic on illustrative inputs shows it: at a 16.5-day median, about 30 pitches sit open at any moment, while the reading itself takes under three hours a business day. Shortening the wait means shortening the queue, and that means setting a clock on every pitch, seeing it before it runs out, and having the answer drafted.

A sender who hears nothing draws a conclusion. As one deal adviser told the 2024 panel, silence "doesn't mean they don't like you; it just means they haven't been able to get to you yet".1 A founder cannot tell the difference, and in the time you take, a rival may reply. We found no independent survey of first-reply times, so this section works with arithmetic you can check, using illustrative inputs. It rests on Little's law, a standard queueing rule: the number waiting equals the arrival rate times the wait.

Counting business days without an argument

Write down how you count. A pitch that arrives on September 22, 2026 with a 10-business-day promise is due on October 6: count the 23rd as day 1 and skip the weekends. If it gets its reply on October 7 it is one day late, and the letter should say so.

Decide whether you count your holidays or the sender's, since Golden Week in China and August in Europe both shift what a reasonable answer looks like. Report the share answered on time alongside the median: the median hides the long tail, and the tail is where the angry emails come from.

A reply promise has a second clock inside it. Once a pitch is routed, the reviewer owes a view, perhaps in 5 business days. Separate the two. If a reviewer has a pitch for three weeks, the first reply can still go out on day 2, truthfully: thank you, this fits, here is who will look at it and when you will hear. An acknowledgment is an answer, as long as it is honest about what comes next.

What a good first reply says, and what it never says

On SIGA, the two figures in circulation count different things. The opinion's $113 million is the lump-sum expectation damages.19 SIGA's own annual report puts the judgment at about $195 million, including pre-judgment interest to January 15, 2015, and says it paid $217 million in all once post-judgment interest ran.20 One law firm's note calls the whole $195 million expectation damages.40

Why inbound licensing pitches keep growing: record deal value, and half the big upfronts from China

In shortLicensing is near a record. Announced biopharma licensing reached $274.3 billion in 2025 and $221.1 billion through September 2026. China-origin assets are 51% of big pharma deals with $50 million or more up front in 2026 to date. Large companies have a reason to look, and more senders, in more languages and formats, are writing to them.

Definitions matter here, because the headlines disagree. J.P. Morgan's 18 deals in 2026 to date, with upfronts of about $5.8 billion on its chart ($5.7 billion in its text), cover large-cap buyers only.39 A wider count cited by Arnold & Porter, from Reuters, puts all cross-border licensing with Chinese biotechs at a record $137 billion in 2025, against $13.9 billion in 2021.23 A trade report puts the average upfront on China licensing deals at $172 million in 2026, up from $52 million in 2022, and names no data source.42 They measure different things.

For a front door, this means a new sender mix: decks translated from Chinese, companies a reader has never heard of, patents granted in China and pending in the US, and data from trials in one country. It also means the people asking for a reply are moving quickly, since competing buyers are reading the same deck. One benchmark of asset-finding tools, written by a vendor, was built around this problem: it mined mostly non-US assets from non-English sources.43

What big pharma pays up front at each stage

The 2026 medians through September are higher at Phase 2 ($363 million) and Phase 3 ($450 million), on two and three deals, so we chart 2025. J.P. Morgan itself says those 2026 medians "reflect limited activity".39

How inbound triage fails: the opened file, the second answer, the missed bid date

In shortMistakes at the front door are cheap to make and hard to see, because a declined pitch leaves no trace in anyone's results. The common ones are lost pitches, stale criteria and a slow reply. The rare ones, an opened confidential file or a wrong decline on a good asset, can cost far more than the whole team. Nobody publishes error rates, so plot your own.

The wrong no costs most, and nobody counts it

A bad deal shows up in a write-down. A good deal you declined shows up only as a competitor's press release, years later, and no internal report will ever connect the two. The authors of an asset-scouting benchmark say that missing one qualifying asset "can mean losing a top partnering or acquisition opportunity worth billions of US dollars", and they have a product to sell, though the logic holds.43

The honest test is a look back. Each year, take your declines from three years ago, check which went on to a public deal or a Phase 3 trial, and read why you said no. If the reason still stands, good. If it reads as "nobody looked", you have found your cost.

Deal pressure runs both ways. One panelist at the 2024 session warned companies against creating "false competitive tension", and another added "Never lie".1 A reader who records what a banker claims, as a claim, keeps the reviewer from inheriting the urgency.

What AI can do with a pitch deck, what the rules say about it, and where it stops

In shortReading a deck into fields, finding an earlier match, scoring against written criteria and drafting a reply all suit AI, with a person deciding. The rules most people worry about mostly do not reach this work. The real risks are the confidential boundary, a made-up citation, and instructions hidden inside a stranger's file.

Software can already read a slide deck faster than a person and put each fact in a field. What it can do reliably is a narrower question, and public evidence is thin.

The most relevant study we found is a February 2026 benchmark of tools for finding drug assets, written by the company behind the top-scoring tool. Its purpose-built tool scored an F1 of 79.7%, a measure that balances finding every right answer against avoiding wrong ones, against 56.2% for the best general research tool and 26.9% for the lowest, on a test set of 22 query-and-asset pairs.43

That benchmark tests outbound search. Inbound triage is an easier and different job: the document is in front of you, so the test is whether each field matches its page. Nobody has published error rates for it. So measure on your own past pitches, against what your reviewers actually decided.

Which AI rules reach an inbound pitch, as of November 2026

Three limits that need a person, whatever the tool

Who answers a partnering pitch today: S&E lead, area reviewers, BD counsel and BD operations

In shortA small team of well-paid experts spends part of its week as a sorting office. A pitch passes between the S&E lead, a therapeutic-area reviewer, BD counsel and BD operations, usually by email, with a spreadsheet or a CRM (the contact and deal database sales teams use) behind it. Most of the time goes to reading and logging, and the elapsed time goes to waiting.

S&E lead to BD operations: four roles around one pitch

Where an estimated 90 minutes on one inbound pitch go

Four ways teams run the inbound flow: inbox and spreadsheet, CRM, outsourced scouting, partnering software

From an 8:41 am email to a signed reply: TV-3 lands on the fit map on day one

In shortIt is Wednesday, October 7, 2026 at Northwind Bio. Priya Ranvelle, Director of Search & Evaluation, has 16 pitches since September 22. Six AI agents read each one and place it on a map by fit and stage. She confirms each routing and signs each reply, beside the map.

1

Every open pitch on one map, sized by the deal

Each bubble is a pitch. Across is fit against the area's search criteria; up is stage; size is the deal on the table, which is the company's ask or the median of comparable deals; color is the therapeutic area. Faded means closed.

Nine open pitches sit at 65 or more, with $11.2 billion on the table. One, PN-12, sits under 50. Above the map are three not on it yet: Talwick, held for legal, and the two new arrivals.

2

The ring around HM-208 has turned red

Each ring is a reply clock. It fills as the 10-business-day promise is used, turns amber in the last two days and red when late. HM-208 from Hollowmere Therapeutics arrived on September 22 and its first reply was due October 6. It scores 74.

The draft says so: "Our apologies that this reply reaches you a day later than we promised."

3

Priya clicks Read 2 new, and six agents take their turn

TV-3, a small-molecule eye implant lasting 6 months for wet age-related macular degeneration, reached the BD inbox at 8:41 am from Tamsin Vision in Dublin. The Confidentiality Check reads the cover note and file labels: non-confidential. The Pitch Reader takes 15 pages into 11 register fields, each with its page.

The Duplicate Finder checks 1,284 earlier pitches and finds no match above 60%. The Fit Scorer works against the eye care criteria, version 2.

4

TV-3 lands in the high-fit corner at 79

Phase 2 and fit 79 put it in the top right zone: high fit with human data. The deal on the table is $950 million, $70 million up front, from comparable deals. Eleven pitches now sit in the route zone, with $12.8 billion on the table, because WXB-17 landed at 67 as well.

The Router suggests Dr. Elena Vasquez, the eye care reviewer, and the Reply Writer has an acknowledgment drafted to Dr. Orla Fennick.

5

Signed on day one of a ten-day clock

Priya reads the letter and signs it from the map. It thanks Dr. Fennick for the non-confidential overview, says the area is one Northwind is actively looking at and names who will review it. The clock now reads "replied in 1 day", and the audit trail records her signature.

Nothing goes to any sender until she signs.

QH-1187: the molecule from May, back with the data Dr. Chen asked for

The pitch in this guide's opening example is on the same map. Qingmere Biologics in Suzhou sent QH-1187 through the portal on October 5, an antibody blocking two inflammation targets, for moderate-to-severe ulcerative colitis, Phase 1b complete. Six agents read, checked and scored it in 38 seconds on this sample. The Pitch Reader turned its 14 slides into 12 fields: the usual 11 plus company history, because slide 2 says "Rivergate Bio is now Qingmere Biologics".

In May, Dr. Maya Chen wrote: "Interesting dual mechanism. Come back to us with Phase 1b data in ulcerative colitis." The match shows the same two targets and antibody format, the same manufacturing partner and cell line, the rename on slide 2 and the same contact on a new email domain. The Duplicate Finder lists these reasons and leaves the merging to a person. The drafted reply opens: "We remember RGB-07 from our conversation in May."

Holding a confidential attachment unopened, and declining PN-12 with the true reason

In shortTwo answers matter as much as a yes. Talwick Therapeutics sent a file marked confidential, and it stays sealed: only its name, size, fingerprint and time are logged, while a CDA request goes to counsel. Pellucid Neuro's PN-12 scores 49, and its letter says why and what would change the answer.

Talwick's email of October 6 attached "CONFIDENTIAL — Talwick data package v3.pdf". The Confidentiality Check read the cover note and file label, and stopped. The register shows three facts from the email and leaves asset, mechanism and data empty on purpose. The setting behind this is locked by legal.

PN-12 is a preclinical gene therapy for Parkinson's disease from San Diego. It scores 49: 21 of 30 on strategic fit, 10 of 20 on differentiation, 2 of 20 on human evidence, 4 of 10 on modality, 8 of 10 on rights and 4 of 10 on competition. A person approves the letter or sends the pitch to Lena Ortiz, the neuroscience reviewer, anyway. Either choice lands on the audit trail, with the reason.

The screens behind the fit map: opportunity register, reviewer lanes, reply calendar, dashboard and search criteria

In shortBehind the map sit the screens each role lives in. The register is one searchable row per asset. Reviewer lanes show who holds what. The reply calendar shows every promise by the day it falls due. The dashboard shows the funnel. Settings hold each area's criteria, the weights and the clocks.

The year so far: from 16.5 days to 2.3, and 226 pitches that never took a reviewer's time

Search criteria, fit weights and promises you set yourself

Six partnering agents draft and suggest; the S&E lead confirms and signs

In shortEach agent has one job and one rule it never breaks. People hold every decision that reaches the outside world: the S&E lead confirms each routing and signs each reply, BD counsel owns the confidential boundary, and each area's owner approves changes to its criteria.

Priya, Maya, Dana and Sam: who decides what on the sample team

Each step lands on an audit trail with who, what and when, kept for seven years in the sample settings. Roles decide who can see and sign what, with maker and checker kept separate: an agent drafts, a person approves, and a criteria change needs the area owner. Botminds is SOC 2 Type II audited and manages information security to ISO 27001.

Decks, portal forms and banker teasers in; a cited register, fit scores and signed replies out

In shortEvery channel a pitch comes through goes in, along with your earlier decisions and your areas' criteria. A cited register, a map, routed reviews and signed letters come out, and everything connects to the systems your BD team already keeps. Outcome figures here are design targets, measured in the live solution.

Connecting the portal, BD inbox, CRM and document store

Design targets for first reply time and reply promise

What a slow partnering inbox costs you: small staff hours, then lost reputation and the asset you missed

In shortThe hands-on time is modest, about $100,000 a year on illustrative inputs. The costs that matter are harder to see: senders who stop calling, reviewers who read the same molecule twice, a confidential file someone opened, and a good asset you declined without looking. Only you can measure those.

Auditing last quarter's pitches to find your own numbers

Sender goodwill, reviewer attention and a claim you cannot rebut

What faster partnering triage returns: hours on each pitch, and one good asset kept in play

In shortPut in your own numbers. Hours alone repay a system only at high volume. One asset you would otherwise have missed can repay it many times, and the reply promise has value of its own. The calculator keeps hours and kept assets apart, so you can see which case you are making.

Hours, reply time and repeat reads: where partnering automation pays

Work out your return from pitches a year, hours by hand and the worth of one asset

A company with, say, 150 pitches a year and a watchful reader may never recover the hours. That team's case, if it has one, rests on memory and the boundary. Buy on those grounds or wait.

How ready your S&E team is: pitch volume, written criteria, earlier decisions, boundary and sign-off

In shortPick the answer closest to your team today. Eight questions score 0 to 2; the last is for reflection. The meter adds up your score as you go, and each question says why it matters and what a strong answer sounds like. The figures in those answers are examples.

What your S&E score means

What to ask a partnering software vendor or scouting firm about confidentiality, citations and scoring

In shortAsk every vendor and adviser the same questions, us included. The answers show how the boundary is enforced, how a field traces to a page, how the score can be challenged, and what happens to other companies' decks.

A six-week replay of last quarter's pitches, before the next live inbound quarter

In shortStart with pitches already answered, so nothing live is at risk. Plant repeats and marked files, keep the answer key, and see what the tool finds. Your reviewers compare it to what they decided. Decide on evidence by week six.

1. Clear the reuse. Ask counsel whether past non-confidential pitches may be replayed in a restricted workspace.
2. Pick 100 past pitches from all four channels, including repeats, declines and a few that became deals.
3. Plant 10 repeats and 5 marked files, renamed and relabeled, and keep the answer key outside the system.
4. Load your criteria, weights and signers, then run blind.
5. Compare the scores, fields and matches with what your reviewers did, and decide.

Replay scorecard: what to count in a partnering triage pilot

Answers on search and evaluation, non-confidential decks, CDAs and China licensing

In shortThe questions people search for, each answered on its own.