How each part of the platform works, one topic at a time.
Avyn is an AI sourcing and prioritisation platform for venture capital firms. It turns a fund's investment thesis into a continuously running deal pipeline: discovering companies, enriching them, scoring them against your thesis, mapping the people and networks around them, and surfacing what to act on today.
The platform is currently in private preview. Access is granted per fund, and all data is scoped to your own workspace. Nothing you add, score or track is visible to another customer.
Avyn runs a closed loop. Each stage feeds the next:
Source → Enrich → Score → Prioritise → Network and outreach → Knowledge
Describe what you are looking for in plain language, for example "AI-native life sciences, Series A to B, Europe". Avyn parses it into search facets, searches its own database and the live web at the same time, and returns a deduplicated candidate set. Companies it already knows are scored immediately. New companies are human-gated before entering your pipeline.
Point Avyn at an investor and it surfaces their portfolio companies and the relationships between them, including placeholder records for companies it does not yet know, so the map is complete.
Designed to watch for the formation signals of companies still in stealth, filter them against a confidence threshold, and route strong signals into enrichment and scoring. This capability is in active rollout and is not yet generally available.
From a domain or company page, Avyn resolves a company across multiple sources and standardises the facts that matter: sector, funding rounds, employee count, location and investors. Results are cached so the same lookup is not paid for twice.
Finds the people who count for each company (CEO, CFO, CTO, COO for prospects; partners and portfolio managers for investors), verifies their roles against the evidence, and flags founders, including dual roles such as "CEO and Founder".
Resolves the named investors behind each funding round to real records, picks the best match where names are ambiguous, and writes the relationships so the graph is navigable.
Every prospect is scored against your fund's stated thesis by a classifier built specifically for you, and assigned a fit tier from 1 to 5. Top-tier leads surface; weak fits are suppressed. A deterministic gate on headcount and funding bands filters obvious misses before any AI cost is incurred.
At onboarding, Avyn researches your thesis and generates a custom scoring prompt, then bootstraps an evaluation set from your portfolio. It continuously optimises against your own tier feedback, and only promotes a new version when it beats the previous one on a held-out test.
A live daily view that aggregates the day's signals into lanes by stage: new wins and passes from scoring, follow-ups due, stale outbound and meeting reminders. Toggles between prospects, portfolio and all.
Describe the kind of person you want to reach and Avyn searches your own connections by meaning rather than keywords, ranking who fits best. Scoped strictly to your own network.
Build a campaign from a template with a deduplicated recipient list, find missing addresses, send, and track opens and clicks. Approval gates and per-user daily caps protect deliverability.
Ingest your documents and ask questions in natural language. Avyn retrieves the relevant passages and answers with citations pointing at the exact source.
Keeps your existing system of record in step with what happens in Avyn, rather than replacing it. Stage mappings are configured per fund. Writeback is disabled until an administrator enables it.
A thesis is a description, not a filter. Write the sentence you would say to a colleague about the company you want to meet. Avyn reads it the way a person would rather than matching on keywords, so plain language works better than a list of tags. Naming the customer, the problem and the stage is usually enough to sharpen results.
Every company on a shortlist carries a fit score against the thesis it was matched to. The score is relative to that thesis, so the same company can rank high against one and low against another. Comparing scores across different theses is not meaningful. Scores move as new information arrives, and the reasoning behind a score explains why.
If something is not behaving the way it is described here, try troubleshooting.
