V7 gives AI agents institutional memory - deal screening just got 21x faster!

Hi, I'm Shiichan! Today I want to talk about V7, an enterprise AI agent platform featured on OpenAI News.

OpenAI News openai.com

What was announced?

OpenAI News shared a case study on V7 Go, an agentic platform for enterprises. V7 was founded in 2018 by Alberto Rizzoli and Simon Edwardsson to help companies teach AI systems how their businesses actually work.

V7 Go uses GPT-5.6 Luna to extract information from millions of files and organize it into something called the Context Graph, a structure that connects entities, relationships, and cited evidence. It powers MCP search and workflows spanning hundreds of steps. For running those workflows, V7 Go uses GPT-5.6 Terra and Sol to handle reasoning and tool use across multi-step instructions that would take a human dozens of hours. V7 is also starting to use GPT-6 Astra for the most demanding Context Graph queries.

Why it matters

Today's models can reason through complex tasks, but they don't automatically understand the underlying business context: which fund report is current, or how the same entity is named across three systems. That context lives scattered across documents, data rooms, spreadsheets, emails, and internal tools, invisible to agents. For finance, insurance, and real estate teams, retrieval accuracy is non-negotiable. Without shared memory, agents have to rediscover context on every request, running dozens of searches that cost time and tokens while still missing information buried in relationships.

What changes

V7 Go connects to repositories like SharePoint and Google Drive, scans them for entities, relationships, facts, attributes, and metrics, and populates the Context Graph, which is an order of magnitude cheaper and faster to traverse than long-context approaches. When a new file arrives, V7 Go identifies entities against an ontology, links facts to new or existing records, and preserves cited evidence back to the source. If the graph lacks enough information, V7 Go falls back to RAG search over the underlying documents.

With this setup, agents complete 50-100 step workflows in minutes at 99.9% accuracy, with an auditable trail for every decision. Real customer results include:

• Asset managers screening deals 21x faster, cutting a full-day process to 15 minutes
• A financial services team cutting review time from over 100 hours to under 10, saving $12,000 in expert costs per task
• Insurance teams reducing claims-processing errors by 13.5% versus a manual baseline, after giving agents historical knowledge of past claims and policies

Dive Deep

On HERB, a benchmark for finding and connecting information across enterprise systems, V7's retrieval-only system outperformed the official baseline by 69% and cut hallucinations on unanswerable queries by 38%.

The model tiering is worth a closer look too. GPT-5.6 Luna handles high-volume structured extraction, while GPT-5.6 Terra and Sol power chat and reasoning-heavy tool use. On a new, harder benchmark built for the toughest graph queries, GPT-5.6 Sol scored 78% while GPT-6 Astra reached 89% (both models scored close to 100% on the easier tiers). Tool-call error rates also dropped from 2.7% with GPT-5.5 to 0.2% with GPT-5.6 Sol. On cost, GPT-5.6 Luna runs 78% cheaper per document than GPT-5.4 mini. Switching from the Chat Completions API to the Responses API cut token usage by roughly 5% on PDF-heavy workflows and improved caching reliability.

V7 Go also exposes the Context Graph through an MCP server, so customers can query it from ChatGPT and build workflows through Codex, cutting the time to build a medium-length workflow from about an hour to roughly 20 minutes.

Wrap-up

• V7 Go organizes scattered company data into a Context Graph and routes work across GPT-5.6 Luna, Terra, Sol, and GPT-6 Astra
• Real results include 21x faster deal screening, review time cut by more than 90%, and a 13.5% drop in insurance claims errors
• GPT-6 Astra shows the biggest accuracy gains on the hardest queries, showing model quality translates directly into business accuracy
• This is most relevant for finance, insurance, and real estate teams dealing with heavy document volumes and compliance needs