Entriva.ai is my new development focus: a program intended to find customers from signals and sell services through autonomous agents.
The working question is whether the commercial process can operate without human involvement, from identifying a possible opportunity through to a purchase.
Where the question came from
Buyers can increasingly use their own AI to investigate needs, evaluate alternatives and assess suppliers before the selling side knows that an opportunity exists.
In my own software purchases, the need for a human intermediary has already fallen away. We identify which tools the work requires and buy them through the process my AI and I are running.
I expect that development to reduce the demand for many human consultative activities. It changes which commercial functions I want to invest in.
MAS.Ai has therefore moved into delivery for existing paying customers only. Further MAS.Ai development stops, and its human consultative side will be discontinued or replaced by AI and robotic systems.
Entriva.ai carries the new working problem.
Current working model
Signal → possible customer need → service fit → proactive outreach or buyer-agent inquiry → commercial dialogue → purchase → outcome → learning.
The program needs to understand the services it represents, which needs they can address and what the supplier can actually deliver.
A signal is a starting point for investigation. The system must establish whether there is a useful commercial opportunity before treating it as demand.
Proactive agents
The proactive agents look for potential customers.
They investigate signals and possible needs, identify a relevant offer, initiate contact and continue the sales process toward a purchase.
The target is autonomous execution across that process. A person should not have to research every account, approve every contact, write every response or take over to close each sale.
Defensive agents
The defensive agents meet potential customers who arrive through their own buying process, including customers represented by AI agents.
They answer questions about the supplier and its services, provide evidence, clarify scope and commercial terms, and work toward a purchase.
Defensive describes their commercial role: representing the supplier when the buyer initiates the assessment.
This requires a supplier agent to participate in the buyer's due diligence and communicate with the buyer's AI.
One commercial understanding
Both agent types need a consistent understanding of what is sold, who it serves, what evidence supports the offer and which commercial terms apply.
Their starting points differ. One finds a potential customer from market signals; the other meets a buyer who has already begun investigating.
Both must be able to complete their part of the commercial process without a human intermediary.
What we need to demonstrate
- Signals can lead to relevant customer and need hypotheses.
- The program can distinguish supplier fit from actual buying intent.
- Proactive agents can initiate and sustain useful commercial dialogue.
- Defensive agents can answer a buyer's AI with accurate, verifiable information.
- The dialogue can progress to a genuine purchase without human involvement.
- Outcomes can improve the system's later decisions.
- The commercial value justifies the cost of running the system.
The intended result is completed sales. Signals, matches and conversations are steps to examine along the way.
Current status
Entriva.ai remains an active working problem. Complete autonomous selling is the development target, not a capability already proven.
A first research foundation for legal-service demand has now been developed. It is an analytical map of where potential demand might originate and what public signs might make those situations worth investigating. It does not yet constitute a validated monitoring engine or evidence of purchased legal services.
What has been built so far
Across 18 practice areas and 390 legal specialties, we described 1,950 potential legal-need hypotheses and 5,842 candidate links between public market signals and those hypotheses. The signal links are not unique events, and they have not been verified empirically.
We also defined a flexible company-first investigation process: begin with a list of companies; examine each company's own public digital channels; check the wider news and press-release picture; and then examine relevant public registers, regulatory developments and political decisions. An interesting external finding should trigger another look at the company's own channels.
Some hypothesized needs have no credible public early warning. A company may already have legal advisers by the time a signal appears. We keep these limitations explicit.
This work is documented as an evolving research asset under Built: Entriva.ai – Legal needs and market-signal map. A field note records the method and what changed.
The next steps are to map concrete candidate sources, test observability and timing against actual companies, and then demonstrate an end-to-end commercial process in a defined use case.
What remains uncertain
How often will publicly observable signals correspond to a real, addressable legal need? Which sources are available for reliable, lawful and economical monitoring, and how early can they inform a decision? How should an agent revise a need hypothesis when the customer disagrees? Which information will buyer agents require before they accept a supplier as a credible choice? Can an agent sustain useful commercial dialogue and complete a purchase at an acceptable operating cost?
Those are the questions Entriva.ai is intended to answer.