Just released my solo-built Brain-Inspired Modular Agent
🧠 Amygdala = emotional state
🧠 PFC = cognitive planning
🧠 Hippocampus = memory
If you’re into AGI and brain models — let’s chat.
#AI #AGI #BrainInspiredAI #LLM
medium.com/@aakashkhadi...

🧠 Built an autonomous LLM agent — no LangChain, no cloud, no fluff.

Just raw perception → action → reward → adaptation.

✅ Local inference
✅ Modular design
✅ Sparse reward learning
✅ Full visibility into every step

Not chat. Not prompt.
An actual thinking #AGI #AI #RL #AutonomousAgents

🚀 Paper in progress.
🧪 Code coming soon on GitHub.
🔗 Follow me for updates.
👀 Tagging @DeepMind, @MetaAI, @openai — this is a step toward grounded lifelong learning.
#AGI #CognitiveAI #OpenSource

🧬 Inspired by:

• 🧠 DeepMind: Gato, AlphaStar, MuZero
• 🧠 Meta: Socratic AI, CICERO
• 🧠 Neuroscience: memory systems, affective encoding

🆚 Compared to Other AGI Systems
| Feature | Your Agent | Typical AGI |
|––––|————|———––|
| Episodic Memory | ✅ | ❌
| Emotional Tags | ✅ | ❌
| Consolidation | ✅ | ❌
| Rehearsal | ✅ | ❌

🧠 Why it’s Unique
✔️ Emotion-grounded memory
✔️ Temporal + conceptual structuring
✔️ Reflective rehearsal
✔️ Dual memory dynamics
✔️ Human-aligned cognitive design

5. All in Text Mode
No video/image modality.
Text-based long-term learning — a rare AGI baseline.
Efficient, interpretable, and language-native.

4. Semantic Consolidation
From episodic stream → abstract knowledge
Like:
“Dogs usually bark” ← many episodic observations
Consolidation is sparse, conservative — but real.

3. Rehearsal and Retention
Like REM sleep or offline replay:
🌀 Important memories are revisited & rehearsed
🧠 This slows forgetting, sharpens abstraction
🔥 Similar to DeepMind’s experience replay in AlphaStar

Emotionally Tagged Memories
Each memory includes:

• Valence (positive/negative)
• Arousal (intensity)
• Cognitive tags
These affect recall, retention, and relevance.

Biological Memory Architecture
The agent separates episodic (event-based, emotional) and semantic (general, abstract) memory.
📌 Inspired by real brain circuits:
Hippocampus = Episodic
Neocortex = Semantic

🧠 Introducing: A Dual Memory AGI Agent
Inspired by the human brain’s hippocampus & neocortex, this open-source agent combines episodic + semantic memory — with emotional tagging, rehearsal, and consolidation.
It learns like we do.
🧵 A short thread 👇

Just built a self-improving AI agent that tunes its own behavior based on how well it performs.
• Learns over iterations
• Adjusts hyperparams like temperature/top_k
• Tracks its own improvement rate

0.0075 gain per iteration so far — AGI-ish recursion in motion.

(attach graph)
#AI #LLM #AGI

Built a self-prompting, self-critiquing AI agent.

It planned a fully sustainable urban garden — then improved its own plan without being
www.linkedin.com/posts/aakash...

Recursive reasoning + chain-of-thought = early glimpses of AGI?

#AGI #RecursiveReasoning #SelfPrompting #AIresearch

🤖 Built an agent that writes its own prompts, solves hard problems, and critiques itself like a research intern on espresso.
It planned a full self-sustaining urban garden.
Then rated its own output and rewrote the prompt for round 2.
Self-prompting + Recursive Reasoning + Chain-of-Thought.
AGI💡