Artificial Intelligence Corporation (AICorp)
The business, the turning points, and the decisions behind them.
01 / WHAT IT BUILT
What the company did
Artificial Intelligence Corporation (later AICorp) was founded in 1975 by Dartmouth natural-language researcher Larry Harris in Massachusetts. Its principal product, INTELLECT, entered the market around 1980 and let non-technical users query mainframe business databases in English. AICorp sold early commercial AI software including INTELLECT, a natural-language database access product. In 1992 it merged with Aion to form Trinzic, which initially continued offering INTELLECT. The merger does not prove the product ended in 1992.
02 / Artificial Intelligence Corporation (AICorp) STANDALONE BUSINESS OUTCOME
Business changes, outcome and evidenced drivers
1980s expert systems and natural-language database products demanded costly integration and maintenance across heterogeneous data structures. The enterprise software market shifted toward client-server tools and broader business-process automation. Merger and strategic repositioning are documented; financial collapse is not established.
The specialist AI tools market weakened as buyers favoured integrated enterprise products and general-purpose tools.
Natural-language database interfaces required schema-, vocabulary- and query-specific customization.
The Aion merger created Trinzic with a broader product portfolio beyond narrowly branded AI tools.
The corporate merger is documented; INTELLECT's exact eventual end date is not established.
03 / EVENT HISTORY
Events in the structured record
- AICorp founded
Larry Harris founded Artificial Intelligence Corporation to commercialise natural-language database technology.
- INTELLECT commercialised
INTELLECT became a representative commercial natural-language database interface; decade is approximate.
- Corporate expansion
AICorp acquired 1st Class Expert Systems and undertook financing.
- Merged into Trinzic
AICorp merged with Aion to form Trinzic; INTELLECT remained in the initial product lineup.
04 / CONTEMPORARY REBUILD ASSESSMENT
Difficulty of rebuilding this project today
A rules-based scenario score drawn from the recorded business model and case information, not a fresh independent company investigation, investment advice or a prediction of success. Review the assessed scope and validation gates below.
LLMs reduce language-interface cost, while enterprise data correctness remains specialised engineering.
- Assessed scopeThe end of AICorp's independent status and the fate of its INTELLECT natural-language database product, not a bankruptcy finding.
- Technical delivery constraintsLLMs reduce language-interface cost, while enterprise data correctness remains specialised engineering.
- Minimum viable testStart with read-only enterprise data QA, validate generated SQL and measure correctness and maintenance.
Reusable query connectors and schema safeguards offer potential scale.
- Reusable scope and scale constraintsReusable query connectors and schema safeguards offer potential scale.
- Validation before scalingStart with read-only enterprise data QA, validate generated SQL and measure correctness and maintenance.
Overall assessment: Start with read-only enterprise data QA, validate generated SQL and measure correctness and maintenance. Scores are scenario estimates of a 2026 rebuild, not historical financial facts or success probabilities.
Assessment date · 2026-11-15
05 / CAUSAL CHAIN
How the factors connect
- 01
Natural-language database access remains valuable, but semantic correctness and integration dominate delivery difficulty.
- 02
A merger or rebranding is not proof of product failure; closure needs separate evidence.
- 03
Modern NL2SQL requires schema controls, permissions, executable validation and cost measurement.