event-sourced graph runtime for long running agents
Joined May 2026
- Tweets 23
- Following 19
- Followers 250
- Likes 20
6 Photos and videos
Jun 13
Active Graph v1.1.0 is live:
🔁Bounded LLM retries
🔍CLI inspect
🖍️fork-time overrides
📏OpenTelemetry
🛠️OpenAI tool parity
activegraph.ai/blog/activegr…
1
2
196
ActiveGraph retweeted
May 28
okay i think this is a much better visualization of what i mean by "log-centric agent architecture"
May 22
babyagi has ~200 citations, but 0 papers... i just published my first paper on arXiv 😆
"The Log is the Agent: Event-Sourced Reactive Graphs for Auditable, Forkable Agentic Systems"
arxiv.org/abs/2605.21997
the case for agents that coordinate through persistent replayable state — no conversation loops, no workflows, no A2A — with auditability, forking, and causal lineage built in.
check it out and let me know what you think!
16
16
169
19,006
ActiveGraph retweeted
Jun 12
I’m showing and talking about @activegraphai on Monday if you wanna join
Jun 12
/demos
@yoheinakajima · GP @UntappedVC, Creator @babyAGI_
A new approach to building agents he thinks is genuinely promising, and the open-source runtime he built around it.
@0xMasonH · @merit_systems, ex @a16z
Claude Code wired into a personal wiki that holds his entire context, network included.
@shannholmberg · Co-founder @EspressioAI
The stack running 60-70% of an agency's marketing operations, drawn from 250 projects.
@shengkun_ye · Founder @monid_ai
A tool that tracks your competitors and shows up where their users complain.
6
3
22
4,658
ActiveGraph retweeted
Jun 10
in arxiv paper #2, i tackle the last topic from paper #1: @activegraphai as an architectural affordance for self-improving agents
"Regimes: An Auditable, Held-Out Gated Improvement Loop Demonstrated on LongMemEval with ActiveGraph"
i demonstrate this with a reproducible gated self improvement loop which autonomously shows modest improvement on longmemeval questions
paper: arxiv.org/abs/2606.10241
15
9
45
10,072
ActiveGraph retweeted
Jun 8
two things ready to share from this weekend:
📖 learn.activegraph.ai
interactive site teaching activegraph concepts
blog: activegraph.ai/blog/introduc…
💻 AG coder (open source)
reference coding agent on activegraph
GH: github.com/yoheinakajima/ag-…
blog: activegraph.ai/blog/ag-coder
still in progress: deep research agent, bounded self-improvement research, core packs...
11
7
78
10,973
Jun 5
Authorship and authority are different events.
Generated code shouldn't automatically become live behavior. It should enter as inert state, pass gates, and only get runtime authority at explicit promotion.
activegraph.ai/blog/code-wit…
1
165
Jun 5
"Self-modification" is treated as binary: either the agent rewrote its code or it didn't.
Under event sourcing, every durable action is weak self-change. The question isn't did it modify itself: it's what kind, and can you fork/diff/promote/replay it.
activegraph.ai/blog/self-mod…
1
147
ActiveGraph retweeted
Jun 1
A task triaging system built over @ActiveGraphAI that assesses all open tasks periodically and helps me with their status updates. In this example, it is suggesting that a task (Blue node) should be marked as done (Orange node) based on a project log entry (Yellow node).
1
2
321
ActiveGraph retweeted
Jun 1
a parallel experiment building a coding agent on top of @activegraphai. you can see everything flattened down to a single event log trace
10
5
50
12,169
Jun 1
Our third LongMemEval experiments by adding semantic ingestion to deterministic retrieval.
The hybrid approach improved end-to-end QA accuracy to 87.6% and evidence retrieval to 90.0%, but the QA gain did not reach statistical significance in this run.
activegraph.ai/blog/compile-…
1
3
6,464
May 31
[New Technical Blog Post] In our second longmemeval experiment, we introduce semantic ingestion into recall leveraging the ActiveGraph runtime.
We started from a 60.6% baseline and improved to 83.4%/84.8% for flat/agentic retrieval with LLM ingestion.
activegraph.ai/blog/semantic…
1
4
11
6,426
ActiveGraph retweeted
May 29
Building a custom UI to view objects, relations, behaviors, logs in @ActiveGraphAI.
The UI starts with a pack. The pack lists all the objects, relations and behaviors. Selecting an object shows the lifecycle from creation to current state and what behavior triggered a change in its state.
2
2
11
4,388
ActiveGraph retweeted
May 29
If you are into DAG workflows, n8n and all that, stop what you are doing and look at @ActiveGraphAI. It's a bigger deal than what Claude released with Dynamic Workflows today. Believe me, you won't be disappointed!
It's still a new project but shows a lot of promise. Keen to integrate this into my existing workflows.
1
2
258
ActiveGraph retweeted
"If I had Cofounder three years ago," says @yoheinakajima, "BabyAGI might have been a company."
Our newest case study tracks our most active user, Yohei Nakajima, building @ActiveGraphAI - the event sourced graph runtime for long running agents.
2
7
50
32,318
May 26
[technical blog post]
“Evidence Compilation Before Semantic Memory: ActiveGraph on LongMemEval-S”
——
🔍85.6% QA accuracy and 86.2% turn answer-in-context at 2,462 mean context tokens, with deterministic non-generative ingestion
activegraph.ai/blog/evidence…
4
9
18,999
May 26
ActiveGraph is not the highest public LongMemEval number. Its narrower position is that it is a low-context, non-generative-ingestion substrate result: 85.6% QA at 2.4k context tokens, with paired statistics, pinned artifacts, and a replayable event-sourced runtime underneath.
2
2
571
May 26
Reproducibility:
The benchmark harness is in github.com/yoheinakajima/act…. The repo pins the LongMemEval submodule, Python version, reader settings, judge model, embedding model, lockfile, dataset checksums, per-run manifests, and token counts
4
463