Should your AI change your records, or only read them? Start with read access only.

With read access, the AI answers questions about your sales records and writes reports from them. A sales leader can ask which deals are waiting on a contract and get the list back in plain M (the system where your team tracks deals) stays exactly as it is.

Write access comes later, in stages, starting with the changes that carry the lowest risk.

If answers and reports are all you want, read access only is a fine place to stop, right? Plenty of teams get what they need there.

Happy to show you how we set it up on a screen share.
0
0
0
11
The first task you give an AI employee should be small enough to review in a few minutes and frequent enough to save time every week.

First, choose a task that repeats on a schedule, such as a daily report or a weekly follow up list. Second, choose a task where the information a system the AI can read. Third, write down what a finished result looks like, so you can check each one.

Start with one task and review the results for a few weeks. Add the next task once the first one runs well, and have a person review anything involving customers or money.

Sometimes a spreadsheet formula or a prebuilt tool covers the task, and I'll tell you when it does, right?

The @phosailabs team is happy to look at your list of recurring tasks, and you can DM me "first task" to get started.
1
0
2
104
Muse is waaayy too proactive
0
0
1
178
If your team repeats a task with a lot of steps every week, an AI agent might be able to run it. Bring the process to Booth 3 at Cancun Tech Day 2026 on November 11, and @phosailabs will tell you whether AI fits or whether software you can already buy covers it.
0
0
1
81
Cancun Tech Day 2026 is five days away, with more than 800 people expected on November 11 at Universidad Anáhuac Cancun.

At Booth 3, @LowCodeAgency can talk with you about super custom software and internal tools for midsize companies. Every project starts with a planning phase oftware belongs to your company, and we stay on as your long-term partner.
1
0
0
85
At LOW/CODE Agency, every equipment tracking project begins with your people. We map how requests, approvals, and returns happen today, then build the system around that. Strategy first, then systems, like pouring the foundation before adding floors.

A partner builds what you'll elf tool fits, use it. A shared sheet and a good habit can carry a small team a long way. Custom software pays off once your approvals and departments follow rules of their own.

Running a company past $5M in revenue? DM me and I will walk you through what fits your team.
0
0
0
59
The voice in this video is an AI clone of mine. I didn't record it.

If you've been on a podcast, someone can clone yours too. So when "you" call finance asking for a wire, what happens?

Two rules: call back on a number you already have, and agree on a code word.
0
0
0
54
AI automation works, and it is also engineering, so a good plan for failure belongs in every proposal you receive.

Here is what to ask a vendor to show you:

•What happens to the process when step 37 fails
•Which risky steps get a second pass check
•Whether every step is ything that reaches a customer

Sometimes plain automation covers the whole job, and I'll tell you so on the call. When something super custom makes sense, we start with Discovery. It's 7,500 bucks and takes two to three weeks. We map your processes with your team, and our tech lead estimates the build. At the end, you can see which steps need judgment, which steps get checks, how long the build takes, and what it costs.

Send me a message with the process you're thinking about, and I'll help you figure out the right fit.
0
0
0
58
Some AI products can ask for the details they need, so a settings screen for every detail may be unnecessary.

In the AI employee product we're designing at @phosailabs, we're removing the delivery settings from scheduled tasks. You write the delivery instructions in the task uses the tools you've connected to send the result. If the right tool isn't connected, the result stays in the chat.

Fewer settings mean fewer screens to learn, and a new user can start with one sentence describing the task. Some settings still need to exist, such as permissions and access, because they protect your data.

I don't know yet how the approach performs for every kind of task, and we'll learn more as people use it, right?

I'm happy to show you the direction when you DM me "settings."
1
0
1
136
This is ABSOLUTELY WILD
This report shares our three subjective conclusions about risk:

1. We believe that, as of Oct 2026, the pace of capability growth on well-specified tasks is likely about 2x what it was from Mar 2024 to Oct 2025. We’re highly uncertain and would not be surprised by 1.3x–3x.
0
0
0
167
Automation follows rules with a yes or no answer, and AI handles cases where the answer depends on context, such as reading an email and deciding who should reply. At @phosailabs we use each one where it fits the process. Booth 3, Cancun Tech Day 2026, November 11.
0
0
0
57
Franklin Delano Frith II. HRM. 16 years of Mexico HR operations running on spreadsheets and manual processes.

Seeing the platform come together:

"This is gonna save so much time. So much time. And we're gonna have everything right in our app."

He was right.
0
0
0
61
Every software project at LOW/CODE Agency starts with Discovery, a planning phase of two to three weeks and delivers a clear roadmap.

We have calls with your team, put together wireframes and documentation, and our tech lead estimates the build, so you know what we're building, what it costs. Booth 3, Cancun Tech Day 2026, November 11.
0
0
1
101
HR confirms an exit once, and every connected app hears about it right away.

Many companies run an equipment app and a separate vehicle app, each with its own list of employees. When HR updates one, the other may hear about the exit a week later.

When we build connected apps them to draw user data from the same central list. Equipment retrieval starts and vehicle access ends in the same moment.

One list of people means every department works from the same facts.

DM me and I will walk you through how to connect your tools.
0
0
0
59
Ask an AI the same question twice and you can get two different answers.

It picks each next word from a list of likely ones. We ran one client-count question three times: 318, 304, 297.

Fix: code runs the count, the AI explains the result.
1
0
0
83
A failed step in our agents retries once and then goes to a person, so a single failure stays contained and does not travel through the remaining steps.

First, a step fails. Second, the agent retries the step once, which covers temporary problems such as a slow page. Third, if person takes over with the log in front of them.

A person also approves anything that reaches a customer, because AI output varies from case to case. Steps that need no judgment run as plain automation, since it costs less and runs more reliably.

The result is an agent where every step has a clear owner, which is the model, a plain automation, or a person.

Send me a message if you have a process like that in mind, and we'll figure out who owns each step.
0
0
1
57
People write requests that match the wording on an AI product's input field, so you can shape the work they ask for by choosing the wording.

In the AI employee product at @phosailabs, the text field reads "Assign a task." The wording invites a defined piece of work, and a person en the work completes.

You can check any AI tool your team uses. First, read the text on the input field. Second, write down the kind of request the text invites. Third, rewrite the text so it matches the work you want done.

Small wording changes cost very, very little, and your team's daily requests follow the wording they see, right?

I'm happy to talk it through when you DM me "assign."
2
0
1
162
Come join our Miami chapter!
Dear Miami founders,

Hampton’s going all in on in-person.

I’m building 10-person peer groups in Miami.

Must be founder or CEO and have $3M+ revenue/funding, or $10M+ exit.

We have 1k+ members & do hundreds of irl events.

Interested? Apply: joinhampton.com/join-city?city…
0
0
0
27
Crazy how quickly Glide, bubble, etc have disappeared.

Now They are even competing vs us for the development $ ...

Theyll end up as small, custom development, closed environment firms working with mid enterprise clients...

Hard to see a successful way out for them...after raising 8/9 figures.
0
0
1
217
One of our clients is an employer of record that hires people in Mexico for US companies. We built an AI agent that finds potential customers for them in about 90 steps, from the LinkedIn search to a draft message that their team reviews. Bring a process with many repeated steps Day 2026, November 11.
3
0
1
146