The best signals show that AI runs in production and has created a problem someone must fix. Watch for inference or ML platform job posts, an AI feature shipped to paying customers, several models in use, and new AI governance roles. A mention of AI in hiring tells you almost nothing.
Why does "uses AI" fail as a signal?
Almost every company now qualifies. In 2024, 78% of organisations used AI, up from 55% the year before (Stanford HAI, April 2025). AI terms were in 5.9% of US job posts by late June 2026. The 2022 peak was 3.3% (Indeed, July 2026).
Some functions are saturated. In December 2025, nearly 45% of data and analytics postings mentioned AI. So did 14.9% of marketing and 8.8% of HR postings (Indeed, January 2026). A post that asks for "experience with AI tools" shows chatbot use. It shows no purchase of serving or evaluation software.
An infrastructure purchase needs three conditions:
- A system in production.
- A team that owns it.
- A cost or failure that the team cannot fix with its current tools.
Look for signals that show at least two of the three.
Which signals point to an infrastructure purchase?
We rank these signals highest for an AI infrastructure or developer tools seller. Each line names the products it fits.
Signals of a production system under load:
- Inference, model serving or ML platform job posts. A team owns production models and needs capacity. Fit: serving, GPU scheduling, observability.
- An AI feature shipped to paying customers, seen in changelogs or release notes. Fit: serving, caching, evaluation, guardrails.
- An engineering blog post or talk on GPU cost, latency or model quality. Fit: cost tools, evaluation, routing.
- Several models in use, named in docs or job posts. Fit: gateways, routing, evaluation.
Signals of a new review or a new budget:
- An AI governance, AI security or model risk hire. Fit: access control, audit logs, data protection.
- A data platform rebuild for AI, seen in job posts or migration posts. Fit: vector search, data pipelines, document processing.
- A funding round that names an AI product. Fit: any layer that the product depends on.
Multi-model use deserves attention. Of 100 CIOs in an a16z survey, 37% used five or more models, up from 29% a year earlier (a16z, June 2025). The same report said agentic workflows made model switches cost a lot of engineering time. That company has a routing and evaluation problem, named or not.
How do you read a job post for buying intent?
Score the words that show ownership of a system. Ignore the words that show tool use.
- Find the verb. "Build", "own", "scale" and "operate" point to a team that runs infrastructure. "Use" and "familiarity with" point to a consumer.
- Find the system. Score posts that name inference, model serving, GPU clusters, evaluation, retrieval or fine-tuning. Skip posts that name only a chat or coding assistant.
- Find the pain. Look for "reduce inference cost", "improve latency" or "move from prototype to production". Quote these words in the first message.
- Count the seats. Two or more open roles on one AI platform team mean the team grows faster than its tools.
- Check the reporting line. A role under a CTO or VP of Engineering sits close to the budget. A role in a business unit shows which unit pays.
If a post passes steps 1, 2 and 3, contact the company this week. If it passes only step 2, put it on a watch list.
How do you match the signal to the right company?
Match the exact company name first, then confirm the domain. Namesakes are the main data risk in signal-based outbound.
In our monitoring, our database could not tie 91 signals to exactly one firm of 11 to 1,000 staff (our data, October 2026). Two more, one a hiring signal, matched a namesake. Our fix is to take the top exact-name match across all company sizes, then apply the size filter.
Many AI company names reuse short common words. Before any contact, check that the domain on the job post matches the domain on the record. Our filter rules for hiring, funding and TVL signals are in which buying signals predict a B2B purchase.
How do you weight signals that arrive together?
Weight combinations above single events. One job post can be a backfill. An inference engineer post in the same month as an AI feature launch shows a production system under load.
- Ownership job post plus a shipped AI feature: contact the platform lead this week.
- Engineering post about cost or latency, alone: send the author a specific comparison.
- Governance or AI security hire plus a shipped feature: lead with deployment options and data handling.
- Funding round only: watch the account until a hiring or product signal follows.
- Generic "AI skills" in non-technical posts: ignore.
When should you contact an account?
Contact an account when public evidence shows two things. AI runs in a product today, and a named team owns a problem your product fixes. Write the first line from that evidence, in the company's own words. Check every other account monthly.
To see the companies with these signals now, start with five named buyers, free, each with its dated signal.