CEO & Co-Founder, Questflow
Bob
Best,
First, we make the new tool fit the work - then we change the work to fit the tool.— Benedict Evans, 2023
We make AI workers bundle with human workers as a new paradigm of work.
Other Startups make an AI worker unbundle as a product.
Incumbents make an AI worker a feature.
In the future, AI workers will work along with human workers to accomplish a common goal.
Therefore, we're not aiming to replace humans with AI; rather, we envision humans working synergistically with AI to spur more innovation and breakthroughs.
To empower next-generation teams collaborating with AI workers to achieve more with less.
So, our mission is:
Most importantly, we believe that the most innovative breakthroughs will emerge from human-AI team collaborations, not in isolation.
This system understands your needs, builds a team of AI and human workers for you, and then executes and optimizes tasks as required. We believe not only will open the door for true human-AI collaboration in the workforce but most importantly provide more time and capital for deep work for breakthrough innovation.
While the task is being done within those projects, you can interact with AI workers as a group or as an individual to provide feedback or ask for more information. AI workers will function just like your team members, with all the context of your project and tasks within the team.
Step 4+: Share and Optimize: Share with teams and interact with AI workers together to provide feedback
Tasks will be executed automatically across AI workers and human workers, and everyone from the team can provide feedback to optimize performance along the way.
Step 3: Automate: Automatically Get Tasks Done Cross Platforms and Workers
Based on your requirements and needs, we will form a group comprising selected AI and human workers to accomplish your tasks. The group could be one person with one AI worker, or multiple people with multiple AI workers.
Step 2: Orchestrate: Build a Human-AI Project Group
Tell us what you aim to accomplish, using simple language. The requirement could be anything related to your work, from operations to finance, to daily office work.
Step 1: Quest: Define Your Requirement
You set up a requirement document, pull up a group with various expertise, and get projects done while checking in along the way to make sure the project meets the needs.
Our approach is inspired by how product leaders usually get projects accomplished in teams.
Here are three simple steps to get your task done with AI workers together:
We are making automation smarter.
Essentially, we are creating the "agent of all agents," akin to a "product manager agent" for all other agents, who can orchestrate AI workers and human workers to get tasks done with our Text-to-Automation tool.
It's crafted to help teams delegate and automate tasks to AI workers, enhancing collaboration and communication within workflows.
Introducing Questflow: The collaborative AI automation tool designed for teams to complete tasks across various platforms with AI workers.
Teams need a new kind of tool for collaboration with AI workers.
So how could we truly delegate and outsource our tasks to AI workers, let them run on autopilot, and then we can focus on deep work while hiring fewer people?
In short, the old way of task delegation is the “APIs of APIs” automation approach, it is a rule-based, step-by-step way for individuals to solve copy-pasting problems across platforms. However, API-based automation by definition does not have agent capabilities, API not only can’t have memory, context, and the ability to reason based on various models, but also it can’t communicate and interact with other APIs and human workers like a real human does. Essentially, automation was dumb.
- Difficult to optimize: AI workers will not be able to execute tasks automatically, collect feedback from human workers during the process, and optimize on their own, making it incapable of improving based on context and feedback.
- Difficult to interact: AI workers can’t communicate with other AI workers and human workers to pass data from one to another, making it impossible to accomplish tasks across platforms if the requirement is more complicated and requires collaboration.
- Difficult to orchestrate: AI workers are completely isolated from other AI workers and human workers, making it difficult to translate your requirements into an actual workflow if you need multiple AI and human workers working together.
There are a couple of main roadblocks in the user journey:
Let’s say your requirement is to automatically filter Emails from GMail and if the email is a PR request, extract the contact information, format and save it in an Airtable Database.
- Functional worker: AI workers who are capable of using third-party tools like searching, extracting information from PDF files, classifying information, and more.
- Knowledge worker: AI workers who have specific domain expertise in Finance, Law, Human resources, or other specific domains who are capable of generating text, audio, image, video, and more.
- Action worker: AI workers who can help you actually get tasks done, like sending emails, updating spreadsheets, booking calendars and meetings, etc.
We have defined three main types of AI workers
Delegating tasks to AI workers will not be afterthoughts in the team workflow, but a premise and necessity. However, with the exponential growth number of AI workers in the marketplace, we find it difficult to actually use AI workers to get tasks done in work.
The collaboration tool for the new workflow will be optimized for a maker’s schedule, to achieve more with less.
- The manager's schedule leaves time for shallow work sporadically. The maker's schedule provides large chunks of time for deep work.
- The manager's schedule focuses on check-ins. The maker's schedule is optimized for deep work.
- The manager's schedule is for hiring and chatting first. The maker's schedule is for delegating and automating first.
The collaboration has completely shifted. We will shift from the old paradigm of the manager's schedule to the maker's schedule, from achieving more with more to achieving more with less.
This also means that we need more collaboration between human workers and AI workers, and less between human workers themselves.
We believe AI agents are not just a concept from textbooks, but more like AI workers just like other human workers will be essential in future workflows. If teams are like this and more work is outsourced to AI workers, then we need a new way of growing the business. We are shifting to the automate-first, generate-second, chat-third approach.
Team workflows are evolving by scaling businesses with fewer human workers, but more AI workers.
They are all next-generation businesses with higher revenue per capita in the team. They are achieving this through all the AI tools available to them, basically achieving more with fewer people.
We have seen signals of this. We have seen teams like MidJourney and Heygen generating millions of revenue with only a few team members. We have seen Shopify stores generating cash with 2-3 worker teams. We have seen creators like @levelsio and Lenny from Substack and X who are building million-dollar businesses with one to two people. On the other hand, we have seen a new type of business called AI Automation Agency (AAA) emerging which are making thousands of dollars per month to help teams who are not expert in AI automating their work.
However, with the rapid growth of AI models and agents, with a wide range of open source/ close source models and agents focused on different areas, we are seeing a trend where we can outsource and delegate tasks to AI agents, hire fewer team members, save money & time, and focus on deep work.
This approach leaves team workflows isolated from AI agents.
In the past, AI agents are afterthoughts in the workforce. Because AI agents are not capable enough to be a part of the workflow, the primary way to get work done in the workforce was working with a human worker, and using AI and automation on the side
The collaboration tool for old workflow is optimized for a manager's schedule, to achieve more with more.
Because most of our tasks are still delegated to human workers, it is very common for users to switch back and forth between IMs, Agents, and Automation tools, where IMs are for chatting with co-workers, Agents are for content generation ranging from text, image, audio, and video, and Automation tools for simple task automation across platforms.
Currently, to incorporate AI agents into the team workflow, we are using the chat-first, generate-second, automate-third approach.
Team workflow needs multiple human workers, working across multiple platforms, with multiple AI agents. It is an asynchronous workflow because people are managing authorization and access to different tools based on their expertise, schedules, and time zones. You need to collaborate with each other to achieve a common goal.
Personal workflow usually means you are using all the tools on your own and you have access to everything. It is a completely synchronous workflow because you are doing everything yourself, and you are basically in control. Usually, you are switching between the app that you are working on (Project management, notes, etc.) and the AI tools (MidJourney, ChatGPT, etc.)
First, we need to differentiate between personal workflow and team workflow while using AI agents.
AI Agents are primarily designed for individuals, not teams.
However, there's a crucial aspect to all this Agent-related work:
We've also observed significant product development in the agent space. Tools like LangChain, LlamaIndex, and others have been pivotal in advancing agent technology, making it accessible for everyday use in our work.
Over the past few months, we've witnessed a lot. New frameworks, insightful essays, technical breakthroughs—you name it. We even saw GPT Dev Day and the rollercoaster of events with the CEO of OpenAI being fired and rehired within three days.
Hi,
