Have you tried specifying the emotion? There's an option to do so and if it's left empty it wouldn't surprise me if it defaulted to rng instead of bland.

For the system prompt I used:

> Read this in a calm, clear, and wise audiobook tone.

> Do not rush. Allow the meaning to sink in.

But maybe I should experiment with something more detailed. Do you have any suggestions?

Something like this:

Character Name: Marcus Cole Voice Profile: A bright, agile male voice with a natural upward lift, delivering lines at a brisk, energetic pace. Pitch leans high with spark, volume projects clearly—near-shouting at peaks—to convey urgency and excitement. Speech flows seamlessly, fluently, each word sharply defined, riding a current of dynamic rhythm. Background: Longtime broadcast booth announcer for national television, specializing in live interstitials and public engagement spots. His voice bridges segments, rallies action, and keeps momentum alive—from voter drives to entertainment news. Presence: Late 50s, neatly groomed, dressed in a crisp shirt under studio lights. Moves with practiced ease, eyes locked on the script, energy coiled and ready. Personality: Energetic, precise, inherently engaging. He doesn’t just read—he propels. Behind the speed is intent: to inform fast, to move people to act. Whether it’s “text VOTE to 5703” or a star-studded tease, he makes it feel immediate, vital.

Sharing a session is independent of the terminal emulator itself. Use tmux for that. There are a handful of good terminal emulators. Weztern, alacritty, and kitty are popular. I use. Tiling window manager so I prefer to avoid tabs and use alacritty for that reason.

Sometimes it is just bias but the 2.5 pro had benchmarks showing the degradation (plus they changed the name every time so it was obviously a different ckpt or model).

Can you give a specific example of something that can't be expressed with a flowchart but can be with rowboat.

In theory you could express most things as a flowchart but the complexity of doing that quickly escalates. A customer support bot that goes beyond informational answers might be a good example for something that is hard to express in a flowchart (without exploding complexity), but can be built in Rowboat.

Here is some personal experience: we previously built Coinbase's automated chatbot and we used a flowchart type builder to do that. This was a intent-entity based system that used deep learning models. It started great, but pretty quickly it became a nightmare to manage. To account for the fact that users could ask things out of turn or move across topics every other turn - we added in concepts called jumps - where control could go from one path to another unrelated path of workflow in on hop - which again introduced a lot of maintenance complexity.

The way we see it is that, when we assign a task to another human or a teammate we don't give them a flowchart - we just give them high level instructions. Maybe that should be the standard for building systems with LLMs?

Is this making sense?

Is the high level instruction compiled to a flowchart under the hood? If so maybe a conversational interface is another layer on a flowchart and not an alternative? Overall it makes sense that flowcharts are limiting when they get big, yes. Product looks cool congrats on the launch.

Thanks!

No, the instructions are not compiled into a flowchart under the hood. We use OpenAI’s agent SDK and use handoffs as a mechanism to transfer control between agents.

There are 3 types of agents in Rowboat: 1. Conversational agents are ones which can talk to the user. They can call tools and can choose to handoff control to another agent if needed. 2. Task agents can’t talk to users but can otherwise call tools and do things in a loop - they are internal agents. 3. Pipeline agent is a sequence of task agents (here the transfer of control is deterministic).

For instance, if we build a system for airline customer support, there might be a set of conversational agents each for different high level topics like ticketing, baggage etc. and internally they can use task and pipeline agents as needed.

Does this make sense?

In my testing this model is quite bad and far behind 235b a22b. https://fiction.live/stories/Fiction-liveBench-Sept-12-2025/...

They should give money that can be used on anything instead of specifically for healthcare. That way you can choose to take care of your kids yourself and put that money towards food than having to work and then outsource childcare.

Uh, people have wasted entire lifetimes chasing wild goose. Newton and Einstein both spent the latter halves of their lives :( despite being geniuses.

I think the primary difference would be they didn't waste billions of dollars in their research.

Isaac Newton dedicated over thirty years to the study and practice of alchemy, writing over one million words on the subject. Comparable in scale to his writings on mathematics and physics.

I'd rather GDP be $1B smaller right now if it meant that Newton had spent another 30 years on physics and math.

If you actually read the claude article it says the same things as the cognition article, it just has a different definition of multi-agent.